This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
PROGRESS was a randomized, open-label, parallel-group phase 3 trial comparing lopinavir/ritonavir plus emtricitabine/tenofovir disoproxil fumarate with lopinavir/ritonavir plus raltegravir in participants with human immunodeficiency virus infection. The registered primary endpoints included virologic response at Week 48 and two safety-related measures assessed through Week 96.
| Feature | PROGRESS |
|---|---|
| Trial phase | Phase 3 |
| Condition | Human Immunodeficiency Virus Infection |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 206 |
| Primary endpoints registered | 3 |
| Primary endpoint types | Binary; Time-to-event |
| Lead sponsor | Abbott |
| Sponsor type | Industry |
| Trial status | Completed |
| Start | 2008-07 |
| Primary completion | 2009-11 |
| ClinicalTrials.gov | NCT00711009 |
2. Clinical Question
The principal statistical question was whether a nucleoside-sparing regimen consisting of lopinavir/ritonavir plus raltegravir could provide a Week 48 virologic response rate that was not more than 20 percentage points lower than the response rate with lopinavir/ritonavir plus emtricitabine/tenofovir disoproxil fumarate.
Population
Participants enrolled in the phase 3 randomized study for human immunodeficiency virus infection. The formal efficacy analysis population was the ITT population defined by the registry as all randomized participants who received at least 1 dose of study drug.
Intervention
Lopinavir/ritonavir (LPV/r) plus raltegravir (RAL), a nucleoside-sparing regimen.
Comparator
Lopinavir/ritonavir (LPV/r) plus emtricitabine/tenofovir disoproxil fumarate (FTC/TDF).
Primary question
Was the LPV/r + RAL regimen non-inferior to LPV/r + FTC/TDF with respect to the Week 48 virologic response endpoint?
3. Trial Design
LPV/r + FTC/TDF
- Lopinavir/ritonavir (LPV/r)
- Emtricitabine/tenofovir disoproxil fumarate (FTC/TDF)
LPV/r + RAL
- Lopinavir/ritonavir (LPV/r)
- Raltegravir (RAL)
- Nucleoside-sparing regimen
4. Endpoints
The registry lists three primary endpoints. Their endpoint types are represented as both time-to-event and binary measures. The statistical analysis posted on ClinicalTrials.gov formally analyzes the Week 48 virologic-response endpoint; the ClinicalTrials.gov record does not contain formal statistical analyses for the two Week 96 safety-related primary endpoints.
| Primary endpoint | Time frame | Type | Formal analysis posted |
|---|---|---|---|
| Percentage of Participants Responding (Plasma HIV-1 Ribonucleic Acid [RNA] Levels Less Than 40 Copies/Milliliter [mL]) at Week 48 Based on the Food and Drug Administration (FDA) Time to Loss of Virologic Response (TLOVR) Algorithm | Baseline to Week 48 | Time-to-event | Yes |
| Percentage of Participants With Moderate or Severe Treatment-emergent, Drug-related Adverse Events | Week 96 | Binary | No formal statistical analysis reported |
| Primary Outcome: Percentage of Participants With Potentially Clinically Significant Laboratory Values | Baseline to Week 96 | Binary | No formal statistical analysis reported |
Week 48 virologic response definition
A participant was classified as a responder at the first of 2 consecutive visits with plasma HIV-1 RNA levels below 40 copies/mL. The participant continued to be a responder until one of the registry-defined loss-of-response conditions occurred, including 2 consecutive values greater than or equal to 40 copies/mL or the specified final-measurement condition.
Week 96 treatment-emergent drug-related adverse events
Treatment-emergent adverse events were defined as those occurring after study drug initiation and within 30 days after the last dose of study drug. The registered endpoint presents moderate or severe treatment-emergent, drug-related adverse events occurring in at least 2% of participants in either treatment arm.
Potentially clinically significant laboratory values
The third primary endpoint concerns potentially clinically significant laboratory values from baseline to Week 96. The registry definition specifies that values occurring in at least 2% of participants in either treatment arm are presented.
5. Statistical Methodology
Non-inferiority framework
The formal Week 48 analysis was explicitly a non-inferiority or equivalence analysis. The registry states that the null hypothesis was that the response rate for LPV/r + RAL was more than 20% lower than the response rate for LPV/r + FTC/TDF.
The LPV/r + RAL arm was considered non-inferior because the lower limit of the exact 95% confidence interval for the response-rate difference was greater than or equal to -20%.
This is a fundamentally different question from asking whether the two response rates are statistically different. A superiority test asks whether the data support a difference away from zero. A non-inferiority analysis asks whether the new regimen can be ruled out as being worse than the comparator by more than a prespecified clinically acceptable margin.
Exact binomial method
The registry reports the exact binomial method, normalized here as the exact / Clopper-Pearson binomial approach. This method is useful when estimating a binomial proportion or a difference involving response proportions without relying on a large-sample normal approximation.
Difference in response percentages
The reported effect measure was the difference in the percentage of subjects responding, defined in the registry as the LPV/r + RAL response rate minus the LPV/r + FTC/TDF response rate.
Under this definition, negative values favor the comparator numerically, while positive values favor the LPV/r + RAL arm. For non-inferiority, the critical issue is whether the lower confidence limit crosses the prespecified -20% margin.
Intention-to-treat analysis
The posted analysis used the intent-to-treat (ITT) population, defined as all randomized participants who received at least 1 dose of study drug. This preserves the randomized treatment assignment as the basis of the efficacy comparison for the population specified by the registry analysis.
Power and sample-size planning
The registry states that the planned sample size of 100 participants per treatment group provided 90% power to conclude that the LPV/r + RAL arm was non-inferior to the control arm, based on a non-inferiority margin of -20% and a type I error rate of 0.05.
The observed enrollment was 206, while the planned sample size described in the analysis notes was 100 participants per treatment group. These are distinct quantities: the former describes the enrolled trial population, whereas the latter describes the design assumption used for the stated power calculation.
6. Results: Week 48 Virologic Response
The single formal statistical analysis posted in the ClinicalTrials.gov record concerns the primary Week 48 virologic-response endpoint. The analysis compares LPV/r + RAL with LPV/r + FTC/TDF in the ITT population.
Difference in percentage of participants responding
95% CI: -12.0% to 8.8% · P = 0.850
Difference defined as LPV/r + RAL minus LPV/r + FTC/TDF.
| Analysis feature | Reported result |
|---|---|
| Endpoint | Percentage of participants responding with plasma HIV-1 RNA levels less than 40 copies/mL at Week 48 based on the FDA TLOVR algorithm |
| Analysis population | ITT: all randomized participants who received at least 1 dose of study drug |
| Groups compared | LPV/r + FTC/TDF vs LPV/r + RAL |
| Effect measure | Difference in percentage of subjects responding |
| Direction of difference | LPV/r + RAL minus LPV/r + FTC/TDF |
| Estimate | -1.6% |
| 95% confidence interval | -12.0% to 8.8% |
| Confidence interval | Two-sided, exact |
| P-value | 0.850 |
| Hypothesis type | Non-inferiority or equivalence |
| Method | Exact binomial method |
The estimated response-rate difference was -1.6 percentage points, with the difference calculated as LPV/r + RAL minus LPV/r + FTC/TDF. Thus, the point estimate places the LPV/r + RAL response rate 1.6 percentage points below the comparator response rate in this analysis.
The point estimate is not a statement that exactly 1.6% fewer individuals benefited, nor does it establish that the two regimens have identical response rates. It is an estimate of the difference between the two group-level response percentages under the registered endpoint definition.
The two-sided 95% confidence interval extends from -12.0% to 8.8%. This interval expresses statistical uncertainty around the estimated difference. It includes zero, so the registry-reported analysis does not establish a superiority difference in either direction. More importantly for the registered non-inferiority question, its lower limit of -12.0% is above the prespecified -20% non-inferiority margin.
The P-value of 0.850 should not be interpreted as the probability that the treatments are equivalent, nor as a measure of the size or clinical importance of the observed difference. P-values quantify evidence relative to a specified statistical hypothesis; they do not measure treatment magnitude. In this trial, the non-inferiority conclusion is driven by the relationship between the confidence interval and the prespecified margin rather than by treating a large P-value as proof of equality.
Because the endpoint is defined through the FDA TLOVR algorithm, the response classification incorporates the longitudinal virologic measurements specified by that algorithm. The registry labels the endpoint as time-to-event, even though the reported effect measure is a difference in the percentage classified as responders at Week 48.
Reading the non-inferiority result directly from the confidence interval
The graphical scale above is only a conceptual display of the reported values; the formal decision comes from the numerical confidence interval. Since the lower confidence limit was -12.0%, it did not reach the -20% non-inferiority boundary.
7. Other Primary Endpoints: Registry Results Without Formal Statistical Analysis
ClinicalTrials.gov lists results for the two additional primary endpoints, but the ClinicalTrials.gov record contains no corresponding formal statistical analysis with an effect estimate, confidence interval, and P-value.
Moderate or severe treatment-emergent, drug-related adverse events
Time frame: Week 96. Treatment-emergent events were defined as those occurring after study drug initiation and within 30 days after the last dose. Events occurring in at least 2% of participants in either treatment arm were presented.
Potentially clinically significant laboratory values
Time frame: baseline to Week 96. The registry presents potentially clinically significant laboratory values occurring in at least 2% of participants in either treatment arm.
For a binary safety endpoint, a formal comparison would ordinarily involve arm-specific event proportions and an appropriate measure of between-arm difference or ratio, together with a confidence interval. For the laboratory endpoint, the analysis would depend on how the clinically significant laboratory categories were defined and tabulated. Those formal comparative details are not included in the ClinicalTrials.gov record.
8. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts by treatment arm. They do not provide a formal statistical comparison or additional adverse-event rates, so the safety section is limited to the reported affected and at-risk counts.
| Safety measure | LPV/r + FTC/TDF | LPV/r + RAL |
|---|---|---|
| Serious adverse events | 14/105 | 12/101 |
LPV/r + FTC/TDF
14/105 participants were affected by serious adverse events among the reported at-risk population.
LPV/r + RAL
12/101 participants were affected by serious adverse events among the reported at-risk population.
These counts should not be converted into a formal treatment comparison without the underlying statistical-analysis specification. In particular, the ClinicalTrials.gov record does not provide a confidence interval, P-value, event-time framework, or prespecified hypothesis for serious adverse events.
9. Non-Inferiority: Why the Margin Matters
The central statistical teaching point in PROGRESS is that non-inferiority is not established by obtaining a nonsignificant superiority P-value. The trial prespecified a -20% non-inferiority margin, and the analysis was constructed around the confidence interval for the difference in response rates.
The reported lower confidence limit was -12.0%, which lies above the -20% boundary.
This distinction matters because a wide confidence interval can include both clinically acceptable and clinically unacceptable differences. A nonsignificant superiority test alone cannot rule out a clinically important disadvantage. The non-inferiority margin supplies the explicit threshold against which the uncertainty is judged.
What the -20% margin means
The margin states the largest loss of response percentage that the trial's non-inferiority framework was designed to tolerate while still considering LPV/r + RAL non-inferior. It is therefore a design parameter, not an estimate of the observed treatment difference.
Why the lower confidence limit is decisive
The estimated difference was -1.6%, but statistical inference must account for uncertainty. The lower limit of -12.0% means that the data's uncertainty extended to a difference substantially more negative than the point estimate. Nevertheless, that lower limit remained above the -20% threshold specified by the trial.
10. Exact / Clopper-Pearson Confidence Intervals
The registry reports an exact binomial method for the primary Week 48 analysis. Exact methods are particularly useful for binomial response data because they avoid relying on an approximate normal distribution when estimating uncertainty around proportions.
The resulting 95% interval reported for the response-rate difference was -12.0% to 8.8%.
The term Clopper-Pearson is commonly used for exact confidence intervals for a single binomial proportion. In the ClinicalTrials.gov record, the statistical method is "Exact / Clopper-Pearson binomial," while the reported effect measure is a difference between treatment-group response percentages. The page therefore preserves the registry's method label rather than implying that the interval can be reconstructed from a single proportion alone.
11. Intention-to-Treat Analysis
The primary analysis used an ITT population defined as all randomized participants who received at least 1 dose of study drug. This definition is important because the analysis population is part of the statistical meaning of the reported estimate.
Why randomization matters
Randomization creates the basis for comparing treatment groups without selecting participants after outcomes are known.
Why the population definition matters
The reported -1.6% difference applies to the specified ITT analysis population, not automatically to every participant enrolled under every possible analysis definition.
In a non-inferiority trial, analysis-population considerations can be especially important because departures from randomized treatment, missing assessments, and protocol deviations can affect the estimated difference and the ability to rule out clinically important inferiority. The ClinicalTrials.gov record identifies the ITT definition but does not provide sufficient detail to evaluate additional per-protocol or modified-ITT analyses.
12. Statistical Methods Explained
Why was a non-inferiority design used?
The trial was structured around whether LPV/r + RAL could retain an acceptable level of virologic response relative to LPV/r + FTC/TDF rather than requiring evidence that it produced a greater response rate. The registry specifies a -20% non-inferiority margin and a corresponding null hypothesis that the LPV/r + RAL response rate was more than 20% lower.
Why is non-inferiority judged against the margin rather than simply by the P-value?
A nonsignificant superiority test does not demonstrate that two treatments are sufficiently similar. A study can have inadequate precision to detect a difference while still being compatible with a clinically important disadvantage. The non-inferiority framework instead asks whether the confidence interval excludes differences worse than the prespecified margin.
What does the -1.6% estimate mean?
The estimate means that the Week 48 response percentage for LPV/r + RAL minus the response percentage for LPV/r + FTC/TDF was -1.6 percentage points in the reported ITT analysis. It does not mean that exactly 1.6% of participants experienced a worse outcome, and it does not by itself establish equivalence.
What does the 95% confidence interval of -12.0% to 8.8% mean?
The interval describes uncertainty around the estimated response-rate difference under the statistical method used. It includes zero, so the registry-reported analysis does not establish a superiority difference. For non-inferiority, the more important feature is that the lower limit, -12.0%, remains above -20%.
What does P = 0.850 mean?
The P-value measures the evidence against the specified null hypothesis under the test framework; it does not represent the probability that the treatments are equivalent. It also does not measure the clinical magnitude of the treatment difference. The non-inferiority conclusion should therefore be read from the prespecified margin and confidence interval rather than from the size of the P-value alone.
Why does the FDA TLOVR algorithm matter?
The endpoint is not simply a single laboratory measurement at Week 48. The registry defines response using the first of 2 consecutive visits with plasma HIV-1 RNA below 40 copies/mL and specifies conditions under which response is lost. The algorithm therefore incorporates the participant's longitudinal virologic measurements into the response classification.
Why is the analysis population explicitly defined?
Statistical estimates depend on who is included. Here, the registry defines the ITT population as all randomized participants who received at least 1 dose of study drug. Stating that definition prevents the reported estimate from being silently generalized to a different analysis population.
13. Censoring and the Time-to-Event Character of TLOVR
The registry classifies the Week 48 response endpoint as time-to-event and bases it on the FDA TLOVR algorithm. The registry-reported analysis, however, reports the final treatment comparison as a difference in the percentage of participants responding at Week 48.
This illustrates an important distinction in clinical-trial statistics: an endpoint can be defined using longitudinal or time-to-loss-of-response information while ultimately being summarized as a responder percentage at a fixed assessment time. The exact handling of all censoring and intermediate observations should be taken from the full statistical analysis plan when reconstructing the analysis in detail.
14. Sample Size, Power, and Type I Error
The registry analysis notes state that the planned sample size of 100 participants per treatment group provided 90% power to conclude non-inferiority under a -20% non-inferiority margin and a type I error rate of 0.05.
| Design quantity | Registered value |
|---|---|
| Planned sample size | 100 participants per treatment group |
| Target power | 90% |
| Non-inferiority margin | -20% |
| Type I error rate | 0.05 |
| Observed enrollment | 206 |
Power is a property of the trial design under specified assumptions; it is not the probability that the observed conclusion is correct. Similarly, the type I error rate is a prespecified error-control quantity, not a description of the observed P-value.
15. Results in Statistical Context
Point estimate
The observed response-rate difference was -1.6 percentage points, defined as LPV/r + RAL minus LPV/r + FTC/TDF.
Precision
The exact two-sided 95% confidence interval ranged from -12.0% to 8.8%, showing uncertainty on both sides of zero.
Non-inferiority boundary
The lower confidence limit of -12.0% was above the prespecified -20% non-inferiority margin.
Formal P-value
The reported P-value was 0.850. It should not be interpreted as a probability of equivalence or as a measure of effect size.
Putting these quantities together gives a more complete statistical picture than quoting the P-value alone. The point estimate is close to zero relative to the -20% margin, the confidence interval remains above that margin, and the registry therefore records the LPV/r + RAL arm as non-inferior under the stated framework.
16. What This Result Does — and Does Not — Mean
The formal registry analysis supports the conclusion that LPV/r + RAL was considered non-inferior to LPV/r + FTC/TDF for the registered Week 48 virologic-response endpoint because the lower limit of the exact 95% confidence interval was -12.0%, above the prespecified -20% margin.
Non-inferiority does not mean that the two regimens have exactly identical response rates. It also does not mean that every individual participant would experience the same virologic outcome under either regimen. The result concerns the prespecified population, endpoint definition, analysis method, and margin.
The confidence interval provides information about precision that the point estimate alone cannot provide. Because the interval extends from -12.0% to 8.8%, the data are compatible with differences in either direction within that interval under the stated inferential framework.
The P-value of 0.850 does not quantify treatment equivalence and does not tell us how clinically important a -1.6% difference would be. The non-inferiority conclusion is based on the confidence interval relative to the prespecified margin.
17. Safety and Efficacy Are Different Statistical Questions
The PROGRESS registry record separates the Week 48 virologic-response endpoint from the Week 96 safety endpoints. This is statistically appropriate because efficacy and safety describe different dimensions of treatment evaluation.
| Domain | Registered question | Data posted on ClinicalTrials.gov for formal comparison |
|---|---|---|
| Virologic response | Percentage responding at Week 48 under the FDA TLOVR definition | Exact difference, 95% CI, and P-value |
| Treatment-emergent drug-related adverse events | Percentage with moderate or severe events through Week 96 | Endpoint definition only; no formal statistical analysis reported |
| Laboratory values | Percentage with potentially clinically significant laboratory values from baseline to Week 96 | Endpoint definition only; no formal statistical analysis reported |
| Serious adverse events | Descriptive safety information reported by arm | 14/105 vs 12/101; no formal comparison reported |
A complete clinical interpretation therefore should not collapse the entire trial into the single Week 48 non-inferiority result. The registry contains distinct efficacy and safety endpoints, and the ClinicalTrials.gov record provides different levels of detail for them.
18. Design Topics Not Specified in the registry Analysis Record
Several statistical topics commonly encountered in clinical trials are not supported by the registry-reported PROGRESS data. They should not be inferred from the existence of the Week 48 analysis.
| Topic | What the ClinicalTrials.gov record supports |
|---|---|
| Stratification | No stratification factors are reported. |
| Factorial design | The design is identified as parallel with 2 arms; no factorial structure is reported. |
| Crossover | No crossover information is reported. |
| Interim analysis | No interim-analysis procedure is reported. |
| Multiplicity adjustment | No multiplicity-adjustment procedure is reported. |
| Missing-data imputation | No separate imputation method is reported. |
| Bayesian methods | No Bayesian method is reported. |
| Cox proportional-hazards model | No Cox model is reported for the registry-reported formal analysis. |
| Kaplan-Meier estimation | The endpoint is classified as time-to-event, but no Kaplan-Meier estimate is reported in the formal statistical analysis. |
Keeping these distinctions explicit is important. The presence of a time-to-event endpoint does not justify inserting a Kaplan-Meier curve, Cox model, or hazard ratio when the registry-reported statistical analysis does not report one.
19. Limitations
- Limited formal statistical output: the ClinicalTrials.gov record contains one formal analysis, for the Week 48 virologic-response endpoint.
- Two additional primary endpoints lack formal comparative output: the registry lists Week 96 treatment-emergent drug-related adverse events and potentially clinically significant laboratory values, but the ClinicalTrials.gov record does not provide their effect estimates, confidence intervals, or P-values.
- Endpoint complexity: the Week 48 response endpoint uses the FDA TLOVR algorithm, so it should not be reduced conceptually to an isolated Week 48 laboratory measurement.
- Non-inferiority interpretation: the -20% margin is a prespecified design threshold. It should not be treated as an estimate of the treatment effect or as a general clinical standard outside this trial.
- Analysis-population dependence: the reported estimate applies to the ITT population defined by the registry as all randomized participants who received at least 1 dose of study drug.
- No registry-reported subgroup analyses: the trial data do not provide subgroup-specific treatment effects or interaction tests.
- No registry-reported missing-data analysis: the provided record does not specify how missing observations were handled beyond the endpoint's registered TLOVR definition.
- No registry-reported multiplicity strategy: although the trial has three registered primary endpoints, the ClinicalTrials.gov record does not describe a formal multiplicity-adjustment strategy across them.
- Safety comparisons are descriptive here: serious adverse-event counts are provided by arm, but no formal hypothesis test or confidence interval is posted on ClinicalTrials.gov for that measure.
20. Why This Trial Matters Statistically
PROGRESS is a useful teaching case because its main statistical question is not simply whether one treatment produced a statistically significant difference from another. Instead, it demonstrates how a randomized trial can be designed to determine whether an alternative regimen remains within a prespecified acceptable loss of efficacy.
| Concept | How it appears in PROGRESS |
|---|---|
| Randomization | Participants were randomized to 2 parallel treatment arms. |
| Non-inferiority design | The primary formal analysis tested whether LPV/r + RAL could be considered non-inferior to LPV/r + FTC/TDF. |
| Non-inferiority margin | The registered margin was -20%. |
| Confidence interval | The exact two-sided 95% CI for the response difference was -12.0% to 8.8%. |
| Exact binomial method | The registry reports an exact binomial method, normalized as Exact / Clopper-Pearson binomial. |
| ITT analysis | The formal analysis used all randomized participants who received at least 1 dose of study drug. |
| P-value interpretation | P = 0.850 illustrates why a P-value should not be treated as evidence of equivalence. |
| Time-to-event endpoint | The Week 48 response endpoint is classified as time-to-event and uses the FDA TLOVR algorithm. |
| Safety endpoints | Two additional primary endpoints assess treatment-emergent drug-related adverse events and potentially clinically significant laboratory values. |
21. A Practical Non-Inferiority Reading Strategy
When reading a trial such as PROGRESS, a useful sequence is to identify the estimand, the direction of the treatment difference, the non-inferiority margin, and the confidence interval before looking at the P-value.
For PROGRESS, this sequence gives: Week 48 virologic response in the specified ITT population; LPV/r + RAL minus LPV/r + FTC/TDF; -20% margin; 95% CI -12.0% to 8.8%; and P = 0.850. The lower confidence limit remains above the non-inferiority boundary.
22. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The formal Week 48 analysis estimated a response-rate difference of -1.6%, with an exact two-sided 95% CI of -12.0% to 8.8%. The lower limit was above the prespecified -20% non-inferiority margin, so the LPV/r + RAL arm was considered non-inferior under the registered framework.
Clinical interpretation
The statistical conclusion concerns the registered virologic-response endpoint and its prespecified non-inferiority criterion. It should not be expanded into treatment recommendations or claims about safety, long-term outcomes, or individual patient benefit beyond the data reported.
23. Overall Statistical Interpretation
The PROGRESS Week 48 analysis reported a -1.6% difference in the percentage of participants responding, with a 95% CI of -12.0% to 8.8% and P = 0.850. The analysis used the ITT population and an exact binomial method.
The key non-inferiority result is that the lower confidence limit of -12.0% remained above the prespecified -20% margin. Under the registry's stated decision rule, LPV/r + RAL was therefore considered non-inferior to LPV/r + FTC/TDF for the Week 48 virologic-response endpoint.
The statistical story is consequently more informative than the P-value alone. The point estimate was -1.6 percentage points, the uncertainty interval extended from -12.0% to 8.8%, and the entire confidence interval remained above the non-inferiority threshold. At the same time, the ClinicalTrials.gov record does not provide formal comparative analyses for the two other registered primary endpoints, so those endpoints should be kept separate from the Week 48 efficacy conclusion.
24. Related Tutorials
Learn more about the methods used in this trial:
25. Related Statistical Calculators
26. Sources
- ClinicalTrials.gov: NCT00711009.
- Linked publication: PubMed record for PMID 22730929.
- Linked publication: PubMed record for PMID 22180523.
Continue through the Clinical Biostats statistical tutorials
Explore the statistical concepts behind randomized trials, non-inferiority margins, confidence intervals, categorical endpoints, and time-to-event analyses.
27. Record Summary
PROGRESS provides a focused example of non-inferiority trial methodology. The randomized phase 3 study compared LPV/r + FTC/TDF with LPV/r + RAL in a parallel, unmasked design with 206 enrolled participants. Its principal formal statistical analysis used the ITT population and an exact binomial method to estimate the difference in Week 48 virologic-response percentages. The reported difference was -1.6%, with a two-sided exact 95% confidence interval of -12.0% to 8.8% and P = 0.850.
The most important statistical feature is the relationship between that confidence interval and the prespecified -20% non-inferiority margin. Because the lower confidence limit remained above the margin, the registry's analysis considered LPV/r + RAL non-inferior to LPV/r + FTC/TDF for the registered Week 48 endpoint. The two additional primary endpoints concern Week 96 treatment-emergent drug-related adverse events and potentially clinically significant laboratory values, but the ClinicalTrials.gov record does not provide formal comparative statistical analyses for them. Serious adverse-event counts are reported descriptively as 14/105 for LPV/r + FTC/TDF and 12/101 for LPV/r + RAL.